بخش های مغز در ارتباط با بازیابی موفق و ناموفق از حافظه اپیزودیک کلامی که توسط توجه تقسیم شده آشکار شده است

Which brain regions are implicated when words are retrieved under divided attention, and what does this tell us about attentional and memory processes needed for retrieval? To address these questions we used fMRI to examine brain regions associated with auditory recognition performed under full and divided attention (DA). We asked young adults to encode words presented auditorily under full attention (FA), and following this, asked them to recognize studied words while in the scanner. Attention was divided at retrieval by asking participants to perform either an animacy task to words, or odd-digit identification task to numbers presented visually, concurrently with the recognition task. Retrieval was disrupted significantly by the word-, but not number-based concurrent task. A corresponding decrease in brain activity was observed in right hippocampus, bilateral parietal cortex, and left precuneus, thus demonstrating, for the first time, involvement of these regions in recognition under DA at retrieval. Increases in activation of left prefrontal cortex (PFC), associated with phonological processing, were observed in the word- compared to number-based DA condition. Results suggest that the medial temporal lobe (MTL) and neo-cortical components of retrieval, believed to form the basis of episodic memory traces, are disrupted when phonological processing regions in left PFC are engaged simultaneously by another task. Results also support a component–process model of retrieval which posits that MTL-mediated retrieval does not compete for general cognitive resources but does compete for specific structural representations.

Manipulating attention by having participants engage in two attention-demanding tasks simultaneously can be used to determine which memory processes draw on cognitive resources for their operation, and to incorporate that knowledge into theories of memory. Dividing attention has been shown to affect encoding much more than retrieval, leading researchers to conclude that the former, and not the latter, requires a general attentional system for optimal performance (Anderson, Craik, & Naveh-Benjamin, 1998; Baddeley, Lewis, Eldridge, & Thomson, 1984; Craik, Govoni, Naveh-Benjamin, & Anderson, 1996; Fernandes & Moscovitch, 2000; Naveh-Benjamin, Craik, Guez, & Dori, 1998). In line with these behavioural findings, neuroimaging studies have shown that neural activation in the prefrontal cortex (PFC) is reduced by dividing attention during encoding (Anderson et al., 2000, Fletcher et al., 1995 and Iidaka et al., 2000; Kensinger, Clarke, & Corkin, 2003; Shallice et al., 1994), but not during retrieval (Iidaka et al.). However, recent behavioural research suggests that large disruptions in episodic memory also can occur from divided attention (DA) conditions at retrieval, specifically when the concurrent task uses material, and/or processing, similar to that used in the memory task. For example, a decrement in verbal memory of about 30% from full attention levels is observed when the concurrent task is word-based, whereas an equally demanding digit-based, or picture-based, task produces a decrement of only 10–15% (Fernandes and Moscovitch, 2000, Fernandes and Moscovitch, 2002 and Fernandes and Moscovitch, 2003). Unlike the general effect observed from DA at encoding, which produces an approximate 50% decline in subsequent memory performance, across numerous types of distracting tasks, effects from DA at retrieval depend on the material or type of processing required in the distracting task. In this study, we use fMRI to examine the neural basis of the interference effect at retrieval, and consider the relevance of those findings for theories of memory. It has been proposed that the large interference effects from DA at retrieval occur primarily as a result of competition for neo-cortical representations, which likely code phonology (Fernandes & Moscovitch, 2003; Fernandes, Priselac, & Moscovitch, 2005; Moscovitch, Fernandes, & Troyer, 2001). The present experiment tests this hypothesis and, in addition, highlights other brain regions that may be implicated. Differences in activation between those DA conditions that affect retrieval (i.e. word-based distracting task), and those that do not (digit-based distracting task), are expected in neocortical sites associated with verbal representation (semantic, orthographical and/or phonological), as well as in medial temporal lobe (MTL) and related structures associated with reactivation of the memory trace. An alternative account is that large memory costs at retrieval result from competition for general attentional resources. Such resources, believed to be mediated by dorsolateral PFC, are needed to coordinate the online processing of dual-tasks (Anderson et al., 2000, Craik, 2001, D’Esposito et al., 1995 and Iidaka et al., 2000), which may be more difficult as the similarity increases, between materials, and/or processing, in the distracting and memory tasks. This idea is consistent with Baddeley's (1992) hypothesis that the ability to coordinate concurrent tasks relies on the central executive (CE) of his working memory model, whose operation requires resources mediated by the PFC. If true, greater activation should be observed in dorsolateral PFC when attention is divided than full, and specifically, it should be greater when materials and/or processing requirements in the distracting task are similar (word-based) to that in the memory task, than different (digit-based distracting task). fMRI allows us to test these hypotheses, and to identify the brain regions associated with varying levels of proficiency of auditory recognition, during full and divided attention conditions.

Our results are the first to identify changes in brain activity, under DA at retrieval, that are related to memory performance. Our findings suggest that the concurrent animacy task disrupted memory directly by suppressing or interfering with temporal lobe function, specifically in the hippocampus, and related episodic memory structures in a network that includes the left precuneus, and parietal cortex. The reduction in activity likely occurs because there is competition between the DA animacy and memory tasks primarily for phonological, but also for semantic, representations in anterior and posterior neocortex. These representations either specify the cues needed to recover information from the hippocampus and related structures, or instantiate the retrieved memories, or both. That there was little variation in activation of the regions of PFC known to be implicated in executive functions during retrieval, suggests, in accordance with the component–process model (Fernandes and Moscovitch, 2000, Fernandes and Moscovitch, 2002 and Fernandes and Moscovitch, 2003; Moscovitch, 1992) that the ecphoric (cue-engram interaction, Tulving, 1983) process itself requires few general processing resources, though some of the pre- and post-ecphoric processes draw on them.